Characterization of Phenylpropene O-Methyltransferases from Sweet Basil: Facile Change of Substrate Specificity and Convergent Evolution within a Plant O-Methyltransferase Family

The Plant Cell, Vol. 14, 505–519, February 2002, www.plantcell.org © 2002 American Society of Plant Biologists

Characterization of Phenylpropene

O

-Methyltransferases from Sweet Basil: Facile Change of Substrate Specificity and Convergent Evolution within a Plant

O

-Methyltransferase Family

David R. Gang,

a,1,2

Noa Lavid,

b

Chloe Zubieta,

c

Feng Chen,

a

Till Beuerle,

a

Efraim Lewinsohn,

b

Joseph P. Noel,

c

and Eran Pichersky

a

a

Department of Molecular, Cellular and Developmental Biology, University of Michigan, Ann Arbor, Michigan 48109-1048

b

Aromatic, Medicinal and Spice Crops Unit, Newe Ya’ar Research Center, Agricultural Research Organization, P.O. Box 1021, Ramat Yishay, 30095, Israel

c

Structural Biology Laboratory, The Salk Institute for Biological Studies, 10010 North Torrey Pines Road, La Jolla, California 92037

Some basil varieties are able to convert the phenylpropenes chavicol and eugenol to methylchavicol and methyleu-genol, respectively. Chavicol

O

-methyltransferase (CVOMT) and eugenol

O

-methyltransferase (EOMT) cDNAs were iso-lated from the sweet basil variety EMX-1 using a biochemical genomics approach. These cDNAs encode proteins thatare 90% identical to each other and very similar to several isoflavone

O

-methyltransferases such as IOMT, which cata-lyzes the 4

�

-

O

-methylation of 2,7,4

�

-trihydroxyisoflavanone. On the other hand, CVOMT1 and EOMT1 are related onlydistantly to (iso)eugenol OMT from

Clarkia breweri

, indicating that the eugenol

O

-methylating enzymes in basil and

C.breweri

evolved independently. Transcripts for CVOMT1 and EOMT1 were highly expressed in the peltate glandular tri-chomes on the surface of the young basil leaves. The

CVOMT

1 and

EOMT

1 cDNAs were expressed in

Escherichia coli

,and active proteins were produced. CVOMT1 catalyzed the

O

-methylation of chavicol, and EOMT1 also catalyzed the

O

-methylation of chavicol with equal efficiency to that of CVOMT1, but it was much more efficient in

O

-methylating eu-genol. Molecular modeling, based on the crystal structure of IOMT, suggested that a single amino acid difference wasresponsible for the difference in substrate discrimination between CVOMT1 and EOMT1. This prediction was confirmedby site-directed mutagenesis, in which the appropriate mutants of CVOMT1 (F260S) and EOMT1 (S261F) were producedthat exhibited the opposite substrate preference relative to the respective native enzyme.

INTRODUCTION

The phenylpropenes are a group of small phenolic mole-cules, derived from the general phenylpropanoid pathway,that are key flavoring elements in many important herbs andspices, including peppercorns, cloves, nutmeg, cinnamon,allspice, pimenta, tarragon, and basil (Gildemeister andHoffmann, 1913; Guenther, 1949; Lawrence, 1992). In addi-tion, the phenylpropenes are important components of thedefensive arsenal of plants or function as pollinator attrac-tants. For example, eugenol is well documented to be an in-hibitor of herbivory (Grossman, 1993; Sisk et al., 1996;

Obeng-Ofori and Reichmuth, 1997) as well as a good nema-tocide (Chatterjee et al., 1982; Sangwan et al., 1990), fungi-cide (Karapinar and Aktug, 1987; Adams and Weidenborner,1996), and bacteriocide (Miyao, 1975). In contrast, methyl-eugenol is an important insect pollinator attractant in manyflowers, for pollinating moths and beetles in particular, andis a female pheromone mimic for several fruit flies (Shuklaand Prasad, 1985).

Little is known about the biosynthesis of the phenylpro-penes, such as eugenol, methyleugenol, and methylchavicol(Gang et al., 2001). Conflicting reports have left the questionopen of how the allyl/propenyl side chain of the phenylpro-penes is formed (Manitto et al., 1974, 1975; Klischies et al.,1975; Senanayake et al., 1977). Other steps in the pathway,however, have been investigated further. Two

O

-methyl-transferase (OMT) activities, operationally defined as eu-genol OMT (EOMT) and chavicol OMT (CVOMT) (Figure 1A),identified in crude protein extracts of sweet basil, were able

1

Current address: Department of Plant Sciences, University of Ari-zona, Tucson, Arizona 85721–0036.

2

To whom correspondence should be addressed. E-mail [email protected]; fax 520-621-7186.Article, publication date, and citation information can be found atwww.plantcell.org/cgi/doi/10.1105/tpc.010327.

506 The Plant Cell

to convert eugenol and chavicol, respectively, to methyleu-genol and methylchavicol using

S

-adenosylmethionine (SAM)as the methyl donor (Wang, 1999; Lewinsohn et al., 2000;Gang et al., 2001). The ratio of these two activities was notconstant between different protein extracts from the samebasil line (Wang, 1999; D.R. Gang and E. Pichersky, unpub-lished data). Affinity purification of EOMT and CVOMT activ-ities from basil leaves identified two polypeptides of similarmolecular mass that could not be separated further (Wang,

1999). Furthermore, both of these activities appear to be re-stricted to the peltate glandular trichomes, the specializedstructures on the surface of leaves and stems that are thesite of synthesis and storage of the phenylpropenes insweet basil (Gang et al., 2001). These results indicated thatthe methylation of eugenol and chavicol could be catalyzedby two separate but very similar enzymes or by a singlepolypeptide whose activity is modulated by some other fac-tor to produce the observed differences in activities be-tween extracts.

High-resolution SDS-PAGE indicated that the subunits ofboth EOMT and CVOMT have a molecular mass of

�

40,000kD, comparable to the size of other plant small moleculeOMTs (SMOMTs) (Ibrahim et al., 1998; Wang, 1999). En-zymes belonging to this class include (iso)eugenol OMT(IEMT), which had been purified and cloned from

Clarkiabreweri

(Wang and Pichersky, 1998). Antibodies to IEMT, anenzyme that evolved recently from caffeic acid/catecholOMT (COMT) in

C. breweri

(Wang and Pichersky, 1998),were able to cross-react with COMT from basil but not withthe purified EOMT/CVOMT protein(s) (Wang, 1999), indicat-ing that the latter probably were not evolved directly fromCOMT but could be more related to other plant SMOMTs.

Many other OMTs of the SMOMT class have been identi-fied in plants. These enzymes play critical roles in the bio-synthesis of many classes of compounds required for plantgrowth and plant defense. COMT itself, which catalyzes the

ortho

-methylation of aromatic diols (Figure 1B), may be in-volved in lignification (Ye and Varner, 1995; Maury et al.,1999; Guo et al., 2001; Parvathi et al., 2001) or may haveother physiological functions (Collendavelloo et al., 1981;Pellegrini et al., 1993). Another group of SMOMTs is in-volved in the methylation of

para

-hydroxyl functionalities inthe biosynthesis of small aromatic compounds in plants. Ex-amples from this group include isoflavone OMT (IOMT; Fig-ure 1A), which catalyzes the 4

�

-

O

-methylation of 2,7,4

�

-trihydroxyisoflavanone, and (

�

)-6a-hydroxymaackiain-3-OMT(HMOMT; Figure 1A), both of which are involved in the bio-synthesis of phytoalexins in legumes (Wu et al., 1997; He etal., 1998). On the other hand, caffeoyl-CoA:SAM OMT(CCOMT), which catalyzes the

meta

-

O

-methylation of caf-feoyl-CoA to form feruloyl-CoA and appears to play an im-portant role in the formation of the coniferyl alcohol moietiesthat are precursors for lignification (Schmitt et al., 1991;Martz et al., 1998; Vander Mijnsbrugge et al., 2000), is notclosely related in sequence to the SMOMT class that con-tains COMT, IEMT, IOMT, and HMOMT.

The three-dimensional crystal structures of IOMT andchalcone OMT (ChOMT), another protein belonging to theclass of SMOMTs containing IOMT that is involved in theproduction of 4,4

�

-dihydroxy-2

�

-methoxychalcone, a

nod

gene inducer (Maxwell et al., 1992, 1993), were reported re-cently (Zubieta et al., 2001). The structures of these two en-zymes are the first to be reported for their class of plantOMTs. This major breakthrough now makes it possible todetermine the specific amino acid residues responsible for

Figure 1. Role of Selected SMOMTs in Plant Specialized Metabo-lism.

(A) Simplified biochemical pathway leading to methylchavicol andmethyleugenol in sweet basil and to the methoxylated isoflavonoids2,7-dihydroxy-4�-methoxyisoflavanone and pisatin in legumes.Transformations catalyzed by CVOMT, EOMT, IOMT, and HMOMTare indicated by single arrows. Double arrows indicate multiple con-versions.(B) Selected conversions catalyzed by COMT.

Basil Phenylpropene

O

-Methyltransferases 507

conferring substrate specificity to this class of enzymes andthus to understand the structural differences that allowsome enzymes, such as COMT, to have broad substratespecificities, whereas others, such as IOMT, have very nar-row substrate specificities. Although the overall three-dimensional structures of the crystallized IOMT and ChOMTproteins are very similar, they have markedly different sub-strate preferences (i.e., their substrates, although similar instructure, are not interchangeable). This substrate discrimi-nation is the result of two factors: first, shape selectivity wasdictated by van der Waals interactions that were unique toeach protein because of the specific complement and ar-rangement of aromatic and aliphatic side chains lining theactive site binding pocket; second, efficient substrate bind-ing was achieved by specific hydrogen bonding patterns(Zubieta et al., 2001). Thus, by the process of random muta-tions altering the amino acid residues surrounding the bind-ing pocket, followed by selection, plants have evolvedOMTs with distinct substrate preferences. This result pro-vides a clear evolutionary mechanism whereby plants havebeen able to produce the great variety of specific enzymaticfunctions necessary for the production of the vast array ofspecialized metabolites (i.e., secondary metabolites) foundin the plant kingdom.

To elucidate the biosynthetic pathway of the phenylpro-penes in basil peltate glands, and specifically to characterizethe OMT activities that convert chavicol to methylchavicoland eugenol to methyleugenol with respect to their struc-tural properties and evolutionary origin, we recently pro-duced an expressed sequence tag (EST) library from basilpeltate glandular trichomes (Gang et al., 2001). In this li-brary, we identified two types of OMTs that are very closelyrelated to each other and that are related to IOMT. Here, wereport that these ESTs represent transcripts for CVOMT andEOMT and that these enzymatic functions are encoded byseparate genes that are closely related to each other and toIOMT and less related to IEMT and COMT. In addition, weshow that CVOMT and EOMT are easily interconvertible viaa single amino acid change, attesting to the relative ease bywhich such specific enzymatic activities can evolve in plants,thereby facilitating convergent evolution.

RESULTS

Isolation and Sequence Characterization of Basil CVOMT and EOMT cDNAs

We reported previously the construction of an EST databasefrom peltate glandular trichomes isolated from the leaves ofbasil cv EMX-1 (Gang et al., 2001). A search in this databasefor potential OMTs revealed several cDNAs with varying de-grees of similarity to known CCOMT sequences and oneclone whose sequence was identical to a previously charac-terized COMT cDNA from basil (Wang et al., 1999). In addi-

tion, 19 ESTs (of 1250, or 1.5% of the total), representingtwo closely related types of sequences, showed the highestsimilarity to IOMT (among OMTs whose function is known).Because all of these 19 ESTs (18 of one type, one of theother type) were found to be incomplete, 5

�

rapid amplifica-tion of cDNA ends (Chenchik et al., 1996; Matz et al., 1999)followed by reverse transcriptase–mediated polymerase chainreaction (RT-PCR) was used to obtain full-length cDNAs,and genome walking (Siebert et al., 1995) verified that therapid amplification of cDNA ends products were completeand accurate.

The two types of complete cDNAs, designated EOMT1and CVOMT1 based on the data presented below, werefound to be 90% identical to each other at the protein level.Their evolutionary relationships to a large family of plantOMTs (SMOMTs, also designated in the literature as the

Figure 2. Members of the COMT Superfamily of Plant SMOMTs Fitinto Two Classes Based on Sequence Homology.

Class I contains COMTs and enzymes such as C. breweri IEMT,which evolved recently from this class. Class II contains OMTs thatuse a variety of substrate structures. Arabidopsis thaliana OMT doesnot fit into either class. The sequences of the genes with names inboldface are compared in Figure 3.

508 The Plant Cell

COMT superfamily) that catalyze the formation of somespecialized, or secondary, metabolites are illustrated in Figure2. This family includes genes for COMT, IOMT, HMOMT, andChOMT as well as a number of related genes with noknown function, for example, putative OMTs from severalspecies in the rose family (

Pyrus

and

Prunus

species)(Mbeguie-A-Mbeguie et al., 1997; Suelves and Puigdomenech,1998). In the evolutionary tree presented in Figure 2, there isa clear division between all known COMT sequences withthe

C. breweri

IEMT on one side and the flavone and iso-

flavone OMTs plus the basil EOMT1 and CVOMT1 on theother side.

From examination of the crystal structure of IOMT, His-257 of IOMT was implicated as the residue responsible forthe deprotonation of the aromatic hydroxyl group of thesubstrate, which initiates the Sn2 attack on the methylgroup of SAM (Zubieta et al., 2001). Asp-288 and Glu-318were shown to constrain the basic His-257, both stericallyand by hydrogen bonding, thereby facilitating efficient catal-ysis. These amino acid residues are conserved in CVOMT1

Figure 3. Amino Acid Sequence Alignment Comparing Selected Members of the COMT Superfamily of Plant OMTs.

CbrCOMT, C. breweri COMT; CbrIEMT, C. breweri IEMT; ObaCOMT1, sweet basil COMT1; ObaEOMT1, sweet basil EOMT1; ObaCVOMT1,sweet basil CVOMT1; MsaIOMT, alfalfa IOMT. Box I through box V domains are conserved among plant SMOMTs. §, SAM binding; ©, catalyticresidue; *, difference between EOMT and CVOMT; ̂ , binding pocket residue for CVOMT; �, substrate binding (from IOMT structure); �, substratebinding on other dyad polypeptide active site; i, residues important for the difference in activity between CbrCOMT and CbrIEMT.

Basil Phenylpropene

O

-Methyltransferases 509

and EOMT1 (Figure 3), suggesting that they are the catalyticresidues in these enzymes as well. In addition, several con-sensus sequences were described previously as being sig-nature motifs of the plant SMOMT group of enzymes(Ibrahim et al., 1998). Based on the crystal structure of IOMT(Zubieta et al., 2001), these motifs, also found in CVOMT1and EOMT1 sequences (boxes I to V in Figure 3), play im-portant roles in SAM binding (boxes I, II, and III) and cataly-sis (box V). Box IV was thought to play a role in metalbinding and to help to orient Asp-288 properly in the activesite (Ibrahim et al., 1998), but no bound metal was found inthe structures of IOMT and ChOMT (Zubieta et al., 2001), sothe actual function of this conserved region is not clear.Other residues implicated from the structure of IOMT asforming part of the substrate binding pocket also are con-served in EOMT1 and CVOMT1 (Figure 3).

Tissue-Specific Expression of Basil

CVOMT

1 and

EOMT

1 Genes and Tissue-Specific CVOMT and EOMT Enzymatic Activities

To determine the tissue in which

EOMT

1 and

CVOMT

1 areexpressed, total RNA was purified from peltate glandular tri-chomes isolated from young developing leaves (

�

1.0 cmlong) of basil lines EMX-1 and SW as well as from wholeleaves of the same size grown under identical conditions.These RNA preparations were used in RNA gel blot analysisto detect and quantitate

EOMT

1 and

CVOMT

1 mRNAs. TheEMX-1 basil line produces phenylpropenes that are methyl-ated at the para position (methylchavicol and some methyl-eugenol), and the SW basil line produces the phenylpropeneeugenol that is not methylated at the para position (Gang etal., 2001). Because of the high sequence similarity between

CVOMT

1 and

EOMT

1, a single phenylpropene OMT probewas used for all RNA gel blot analysis, and the signal de-tected in these experiments was the sum of the two types oftranscripts. As can be seen in Figure 4A, lane 2, basil lineEMX-1 peltate glands express these genes at much higherlevels than whole leaf tissue (lane 1), whereas transcripts for

EOMT

1 and

CVOMT

1 are lacking in the SW line (lanes 3 and4). These results verified previous assays that indicatedthat active EOMT1 and CVOMT1 enzymes were restrictedto the peltate glands (Gang et al., 2001) of basil varietiesthat produce

para

-

O

-methylated phenylpropenes.Further experiments were performed to determine how

EOMT

1/

CVOMT

1 gene expression and activity were regu-lated developmentally in basil line EMX-1. Leaves of threesizes—0.5, 1, and 3 cm long—were collected and analyzed(as a whole) for transcript levels and for specific enzymaticactivities. Total RNA was isolated and hybridized with thebasil phenylpropene OMT probe. As can be seen in Figure4A, lanes 5 to 7, respectively, the basil OMTs were ex-pressed at much higher levels in 0.5- and 1-cm leaves thanin 3-cm leaves. EOMT and CVOMT enzyme activities (Fig-ure 4B, lanes 5 to 7) also were much higher in very young

developing leaves (0.5 cm) than in older, more developedleaves.

To further localize the site of EOMT and CVOMT enzy-matic activities and mRNAs in older leaves, 1- and 3-cmleaves were cut perpendicularly to the midvein into halvesand thirds, respectively, and analyzed for basil OMT expres-sion and enzyme activity. As can be seen in Figure 4A, lanes8 to 12, the OMT transcripts are expressed at much higherlevels in the basal portion of the leaf (in both 1- and 3-cmleaves), which is still undergoing cell division and expansion,than in the leaf tip, which has ceased most growth pro-cesses. Enzymatic assays (Figure 4B, lanes 8 to 12) sup-ported this observation as well. In all tissue samplesexamined, the amount of EOMT activity was approximatelyequivalent to CVOMT activity (Figure 4B).

Figure 4. Tissue-Specific Expression of Basil PhenylpropeneOMTs.

(A) RNA gel blot analysis of gene expression. Lanes 1 to 4 (3 �g oftotal RNA per lane) show comparison of gene expression in wholeleaves (lanes 1 and 3) and in isolated peltate glandular trichomes(lanes 2 and 4) from basil lines EMX-1 (lanes 1 and 2; produces para-O-methylated phenylpropenes) and SW (lanes 3 and 4; producesnon-para-O-methylated phenylpropenes). Lanes 5 to 7 (5 �g of totalRNA per lane) show comparison of gene expression between differ-ent stages of leaf development from EMX-1 whole leaves 0.5 cmlong (lane 5), 1 cm long (lane 6), and 3 cm long (lane 7). Lanes 8 to12 (5 �g of total RNA per lane) show comparison of gene expressionin different parts of EMX-1 leaves: lane 8, 1-cm leaf apical half; lane9, 1-cm leaf basal half; lane 10, 3-cm leaf apical third; lane 11, 3-cmleaf middle third; lane 12, 3-cm leaf basal third.(B) Specific activities for EOMT (black bars) and CVOMT (white bars)enzymes. Lane numbers indicate tissues as described for (A). Barsindicate �SE.

510 The Plant Cell

Because our RNA gel blot experiments did not distin-guish between

EOMT

1 and

CVOMT

1 transcripts, we used aquantitative RT-PCR procedure (see Methods) to determinewhat proportion of transcripts observed in the RNA gelblots were

CVOMT

1 transcripts and what proportion were

EOMT

1 transcripts. In leaves of various stages of develop-ment, as well as in isolated peltate glands,

CVOMT

1 ac-counted for

�

90 to 95% of the total transcripts, whereas

EOMT

1 accounted for the rest (data not shown). The ratioof

CVOMT

1 to

EOMT

1 transcripts, together with the obser-vation that EOMT catalytic efficiency with eugenol is

�

10-fold higher than CVOMT catalytic efficiency with chavicol(see below), explain why the levels of EOMT activity in thetissues examined were approximately equivalent to CVOMTactivity levels.

Enzyme Activities of Recombinant Proteins

The coding regions of

CVOMT

1 and

EOMT

1 were trans-ferred to the expression vector pCRT7/CT-TOPO TA (Invi-trogen, Carlsbad, CA) for functional expression as nonfusionproteins. COMT1 from basil (Wang et al., 1999), expressedfrom the pET28 vector, also was characterized.

All three OMT proteins (CVOMT1, EOMT1, and COMT1)were purified to 70 to 95% homogeneity using gel filtrationand ion exchange chromatographic separations. Purifiedenzyme for each recombinant protein was used for the de-termination of general catalytic properties and kinetic pa-rameters and the determination of substrate specificity(Table 1). All three proteins displayed in SDS-PAGE a sub-unit molecular mass of

�

40,000 D (data not shown), whichis very comparable to the values calculated from the trans-lated amino acid sequences of the corresponding genes.This also is the size obtained for the CVOMT/EOMT mixturethat was purified previously from basil leaf extracts (Wang,1999). However, all three recombinant proteins eluted froma calibrated gel filtration column as 81,000- to 87,000-Dproteins (EOMT eluted as slightly smaller than CVOMT andCOMT), suggesting that all three proteins exist as ho-modimers in solution. This is not surprising, because othermembers of this family of OMTs, including the recently crys-

tallized IOMT and ChOMT, also exist in solution as ho-modimers (Zubieta et al., 2001).

The expressed recombinant proteins all possessed a pHoptimum from 7 to 8. In addition, increasing concentrationsof salt (NaCl)

�

50 mM (or ammonium sulfate of comparableionic strength) reduced enzymatic activity linearly with saltconcentration. No difference in activity was observed for theenzymes from 0 to 50 mM NaCl. All three enzymes ap-peared to be catalytically stable at temperatures up to

�

30

�

C. Assays with crude basil leaf extracts for EOMT,CVOMT, and COMT activities also displayed the same pHoptima and temperature dependence, indicating that the re-combinant proteins behaved similarly, if not identically, tothe native enzymes.

All three enzymes (CVOMT1, EOMT1, and COMT1) wereevaluated under standard conditions for their ability to cata-lyze the SAM-dependent

O

-methylation of a large number ofpotential substrates. The results of these assays, as well asa representative sampling of the compounds tested, are il-lustrated in Figure 5. As can be seen, CVOMT1 is the mostspecific of the enzymes in terms of substrate preference,and it catalyzes the

O

-methylation of chavicol much morereadily than it does any other compound tested. Other com-pounds with a

para

-hydroxyl functional group, such as eu-genol,

t

-isoeugenol,

t

-anol, caffeic acid, and phenol, servedas poorer substrates. EOMT1, on the other hand, was foundto methylate eugenol most efficiently, with

t

-isoeugenol andchavicol being poorer substrates. Interestingly, the sub-strate preference of EOMT1 differed from that of

C. breweriIEMT (Wang and Pichersky, 1998, 1999), which instead usest-isoeugenol and eugenol almost equally well as substrates,although the former was a slightly better substrate. Chavicolalso was a poorer substrate for EOMT1 than it was for IEMT.Thus, EOMT1 from basil is more specific in its substratepreference than IEMT from C. breweri.

CVOMT1 and EOMT1 discriminated against potentialsubstrates that did not have an allylic side chain in the paraposition relative to the hydroxyl group that was methylated.Thus, chavicol, but not t-anol, served as a good substratefor CVOMT1, and eugenol served as a better substrate forEOMT1 than did t-isoeugenol. t-Anol and t-isoeugenol pos-sess propenyl and not allylic side chains (Figure 5). In addi-

Table 1. Properties of Basil Phenylpropanoid OMTs

Enzyme

Holoenzyme Mass (D) by Gel Filtration

SubunitMass (D) bySDS-PAGE

Calculated Subunit Mass (D)

pH Optimum

ApparentKm (�M)for SAM Substrate

ApparentKm (�M)for Substrate

ApparentVmax (pkat·mg1) kcat (s1)

kcat/Km

(nM1·s1)

COMT1 87,000 �40,000 39,529 7 to 8 64 Caffeic acid 47 19,500 1.3 102 0.27Catechol 290 6900 4.6 103 0.02

EOMT1 81,000 �40,000 40,237 7 to 8 5 Eugenol 3 4600 3.1 103 1.23t-Isoeugenol 10 1900 1.3 103 0.13Chavicol 7 1300 8.7 104 0.13

CVOMT1 87,000 �40,000 39,937 7 to 8 5 Chavicol 6 1200 8.0 104 0.13

Basil Phenylpropene O-Methyltransferases 511

tion, compounds with hydrophilic functionalities on the endof the alkene side chain (e.g., coniferyl alcohol, ferulic acid,and p-coumaric acid) were poor substrates for bothCVOMT1 and EOMT1.

Basil COMT1 displayed a broad substrate preference, us-ing a number of aromatic diols as substrates. Ortho-diols(e.g., catecol) were much better substrates than were meta-or para-diols (e.g., resorcinol and hydroquinone, respec-tively). Although CVOMT1 and EOMT1 preferred substrateswith very hydrophobic side chains, COMT1 preferred sub-strates with either no alkene side chain (e.g., catecol) or withhydrophilic functionalities on the end of the side chain (e.g.,caffeic acid). However, the aromatic hydroxyl group that be-comes methylated by COMT1 resides at the position metato the alkyl side chain (if such a side chain is present in thesubstrate). Thus, caffeic acid, which possesses a hydroxylgroup meta to the side chain, is methylated, whereas p-cou-maric acid, whose hydroxyl group is in the para position, isnot a substrate for COMT1.

Kinetic parameters for each enzyme also were deter-mined. These are listed in Table 1. The kinetic parametersfor both EOMT1 and CVOMT1 with regard to chavicol as asubstrate were identical, especially the apparent catalyticefficiency (kcat/Km ratio) and the absolute values for apparentVmax and apparent Km. This finding indicates that both en-zymes treat chavicol identically as a substrate (see legendto Figure 5). In addition, for EOMT1, the apparent catalyticefficiency for chavicol and t-isoeugenol was identical and�10-fold lower than that for eugenol, again supporting eu-genol as the favored substrate. However, the apparent Km

for t-isoeugenol was approximately twofold higher than theapparent Km for chavicol, confirming the observation thatbinding of an allylic, and not a propenyl, side chain is favoredby EOMT1. The apparent catalytic efficiency of t-isoeugenolO-methylation was the same as for chavicol because of theincrease in apparent Vmax, presumably attributable to themethoxyl group at the end of the substrate, which would beable to form hydrogen bonds to stabilize the binding of thetransition state at that end of the molecule once it hadbound. EOMT1 and CVOMT1 reacted in a like manner to-ward the cofactor SAM, with an apparent Km for this sub-strate at 5 �M for both enzymes.

The only apparent catalytic difference between EOMT1and CVOMT1 is that CVOMT1 has a very restricted sub-strate preference (see above), whereas EOMT1 can use eu-genol and t-isoeugenol much more effectively as substrates(with much higher apparent Vmax and higher apparent kcat/Km

ratio).COMT1 had much higher Km values (10-fold higher) for

SAM and its substrates than did EOMT1 and CVOMT1 (Ta-ble 1), supporting the notion that it is a less specific enzyme.However, COMT1 is able to achieve a catalytic efficiency to-ward its substrates that is similar to the apparent catalyticefficiency (kcat/Km) of EOMT1 toward its substrates, be-cause COMT1 displays a much higher apparent Vmax. In ad-dition, COMT1 did not appear to display the dramatic

substrate inhibition that was observed for CVOMT1 andEOMT1.

Modeling of Basil EOMT and CVOMT Active Sites

Basil CVOMT1 and EOMT1 tertiary and quaternary struc-tures were modeled based on the recently published crystalstructures of IOMT from alfalfa (Zubieta et al., 2001). Thesemodels supported the observation that these two enzymes

Figure 5. Comparison of Relative Specific Activities of BasilCVOMT1, EOMT1, and COMT1 with a Variety of Substrates.

For each enzyme, the specific activity that is the highest is set at100%; although EOMT has similar activity with chavicol as CVOMT,its activity with eugenol is fourfold higher.

512 The Plant Cell

exist in solution as dimers, with a small N-terminal domaininvolved in dimerization, as was observed for IOMT (Zubietaet al., 2001). Based on the molecular modeling data, EOMT1appears to form a more compact homodimer than CVOMT1,which appears to be more extended. This would explain theslightly different behavior of EOMT1 on the gel filtration col-umn (eluting as an 81,000-D protein instead of 87,000 D, asCVOMT1 does).

Interestingly, the active sites of CVOMT1 and EOMT1(Figures 6A and 6B, respectively) are almost identical. Thesubstrate binding pockets of both enzymes are lined withthe same aromatic and aliphatic amino acid side chains(with one exception; see below), producing a very hydro-phobic environment for substrate binding. The substrateschavicol and eugenol also were modeled into the activesites of their respective enzymes. Based on this modeling,several amino acid residues were implicated in catalysis.

These include Ser-261, His-264, Asp-265, and Glu-323 forEOMT1 and His-263, Asp-264, and Glu-322 for CVOMT1.The amino acid residue numbers are off by one between thetwo enzymes as a result of the presence of Asn-91 inEOMT1, which is lacking in CVOMT1. Asn-91 resides in anexternal loop region of the protein that is far removed fromthe active site and appears to play no role in catalysis.

Based on these models, one major difference was ob-served between the active sites of EOMT1 and CVOMT1. InEOMT1, Ser-261 appears to play a critical role in forming ahydrogen bond with the para-hydroxyl group of the sub-strate eugenol, stabilizing the binding of the substrate in theactive site. His-264 appears to interact with Ser-261 in thisprocess. In addition, His-264 appears to be the catalyticresidue, responsible for initiating the deprotonation of thehydroxyl group to form the reactive substrate intermediatethat attacks the electrophilic methyl group of the SAM co-factor in the nucleophilic addition of the methyl group to thesubstrate. Asp-265 appears to form a hydrogen bond withthe methoxyl oxygen of the eugenol substrate, further stabi-lizing substrate binding. Glu-323, which also is near His-264in the active site, can form a hydrogen bond to the imidazolering of His-264, apparently stabilizing the catalytic process.Thus, the interactions between His-264, Asp-265, Glu-323,and Ser-261 with each other and with the substrate appearto be critical for efficient catalytic function.

In CVOMT1, however, position 260 contains a phenylala-nine residue instead of serine. Phe-260 is not able to formthe hydrogen bond necessary to stabilize the binding of thesubstrate. Instead, this stabilization must be performed byHis-263 alone. In addition, the larger phenyl ring of Phe-260causes a shift in the orientation of His-263, which must bestabilized by a shift in the orientation of Glu-322. The alter-ation in the position of His-263 also causes a shift in the ori-entation of the bound substrate, making hydrogen bonds toAsp-264 less favorable. Thus, substrates such as eugenoland t-isoeugenol are less likely to be stabilized in the activesite. Chavicol, which is smaller and does not possess themeta aromatic methoxyl functional group, still is able tobind.

In addition to the modeling of CVOMT1 and EOMT1, wemodeled the C. breweri IEMT three-dimensional structure(Figure 6C), based on the IOMT structure, and found that itsactive site was very similar to the active site of EOMT1 andCVOMT1. Again, His-273, Asp-274, and Glu-333 appear tobe involved in catalysis and substrate binding, but the serinefound at position 261 of EOMT1 is a tryptophan (Trp-270) inIEMT. This causes lower binding affinity for the substrate (a10-fold increase in apparent Km compared with EOMT1[Wang and Pichersky, 1999]). Unlike CVOMT1, however,IEMT does not appear to display substrate inhibition, so itsapparent catalytic efficiency toward t-isoeugenol (Wang andPichersky, 1999) is very similar to the apparent catalytic effi-ciency of EOMT1 toward eugenol (because of higher turn-over number at high substrate concentration). In addition,Phe-167 of IEMT is shifted away from the binding pocket

Figure 6. Stereo Views of the Three-Dimensional Structures of theActive Sites of Basil CVOMT (A), Basil EOMT (B), and C. breweriIEMT (C) as Determined by Molecular Modeling Based on the Crys-tal Structure of Alfalfa IOMT.

The most efficient phenylpropene substrate for each enzyme isshown bound, as is S-adenosylhomocysteine (the cofactor product),which was cocrystalized with IOMT to produce the original three-dimensional structure.

Basil Phenylpropene O-Methyltransferases 513

compared with Phe-156 of EOMT1 and Phe-155 ofCVOMT1, altering the geometry of the binding pocket (Fig-ure 6). This could explain the broader substrate allowancefor IEMT compared with CVOMT1 and may explain why thisenzyme prefers substrates with propenyl side chains overthose with allylic side chains.

Furthermore, as described previously, IEMT appears tohave evolved recently from COMT in C. breweri (Wang andPichersky, 1999). One of the major changes noted betweenthese two enzymes is that a short sequence motif, MNQ, atresidues 133 to 135 of COMT is changed to TAT in IEMT.This change was found to be responsible for the majority ofthe substrate discrimination observed between these twovery closely related enzymes (Wang and Pichersky, 1999).When we evaluated the modeled IEMT (also based on theIOMT structure; Figure 6C), we found that the small aliphaticresidue Ala-134 is located in the substrate binding pocket.The side chains of Thr-133 and Thr-135 appear to be di-rected away from the binding pocket. Significantly, the sidechain of Ala-134 resides at the end of the binding pocketopposite where the catalytic His-273, Asp-274, and Glu-333are located. In this position, Ala-134 is situated near the endof the phenylpropene side chain of the t-isoeugenol sub-strate, where it can stabilize hydrophobic moieties of poten-tial substrates. In CVOMT1 and EOMT1, the residues thatreside in this location in the active site also are small hy-drophobic residues (Val-121 and Val-122 in CVOMT1 andVal-122 and Val-123 in EOMT1). These amino acid sidechains would not interfere with the binding of alkene groupsbut would discriminate against hydrophilic group binding. IfAla-134 in IEMT were changed to the asparagine found inCOMT, then the very hydrophilic side chain of the aspar-agine would be in close proximity to the very hydrophobicpropenyl chain, destabilizing substrate binding. In COMT,however (or in the IEMT TAT→MNQ mutant, which uses caf-feic acid as the preferred substrate [Wang and Pichersky,1999]), this hydrophilic group could readily form hydrogenbonds to the oxygenated functionalities of the substrate,thereby stabilizing substrate binding.

Construction and Analysis of Mutated CVOMT1 and EOMT1 Enzymes

To determine the contribution to eugenol discrimination ofSer-261 in EOMT1 and of Phe-260 in CVOMT1, which wasdeduced from the modeling of their active sites, we changedthe Ser-261 codon of EOMT into a phenylalanine codon (asingle C-to-T mutation) and changed the Phe-260 codon ofCVOMT into a serine codon (a single T-to-C mutation).CVOMT1 and EOMT1 constructs containing these modifica-tions, CVOMT1 F260S and EOMT1 S261F, respectively,were inserted into the pCRT7/CT-TOPO TA expression vec-tor, sequenced to verify that the correct modifications hadbeen made and that no other modifications were introducedinto the sequences, and expressed in Escherichia coli. The

corresponding mutant recombinant proteins were assayedfor substrate preference with chavicol, eugenol, t-isoeu-genol, and caffeic acid (Figure 7). By comparing the activityof these mutants with the activity of the wild-type enzymes(Figure 5), two conclusions can be drawn. First, theCVOMT1 F260S mutant behaves like the native EOMT1 en-zyme with regard to substrate preference. Instead of dis-criminating against eugenol, as the native protein does(Figure 5), the CVOMT1 F260S mutant uses eugenol muchmore efficiently than chavicol (Figure 7) and is able to cata-lyze the SAM-dependent O-methylation of t-isoeugenol withthe same apparent efficiency as it did chavicol. In contrast,the mutant EOMT1 S261F functions like the native CVOMT1enzyme. Instead of being able to use eugenol, t-isoeugenol,and chavicol, as the native enzyme does (Figure 5), theEOMT1 S261F mutant is able to effectively catalyze only theO-methylation of chavicol (Figure 7).

DISCUSSION

Basil glands express one OMT that is specific for chavicoland another OMT that has similar activity with chavicol but

Figure 7. Substrate Specificity of Mutant Basil PhenylpropeneOMTs.

The relative specific activities of mutant basil CVOMT1 (CVOMT1F260S) and EOMT1 (EOMT1 S261F) are compared with the specificactivities of the corresponding native enzymes (CVOMT1 andEOMT1, respectively) using key substrates: chavicol (white bars),eugenol (black bars), t-isoeugenol (diagonally hatched bars), andcaffeic acid (cross-hatched bars). For each enzyme, the specific ac-tivity that is the highest is set as 100%.

514 The Plant Cell

has much higher activity with eugenol. Preliminary studieshad demonstrated that the ratios of chavicol and eugenolOMT activities were not constant in sweet basil leaves(Wang, 1999; Lewinsohn et al., 2000; Gang et al., 2001). Thiscould be explained by the presence of two separate OMTgenes that were not always expressed at the same level orby the existence of some other factor that could modulateone or more of the OMT activities. If the observed differ-ences were the result of two separate phenylpropene OMTenzymes, it was not clear whether the enzymes would havecompletely separate substrate preferences (i.e., each wouldbe very specific for its substrate) or if the enzymes wouldoverlap in substrate specificities. The cloning and heterolo-gous expression of CVOMT and EOMT from sweet basilpeltate glandular trichomes have answered these questions.

Two separate OMT gene types, which represent CVOMTand EOMT enzymes, were isolated from the peltate glandu-lar trichomes of basil line EMX-1 leaves (Figure 3). RNA gelblot analysis demonstrated that the mRNA transcripts en-coding these enzymes are highly expressed in the peltateglandular trichomes of the basil line (EMX-1) that produceslarge amounts of methylchavicol (and some methyleugenolas well) but are expressed at very low levels in a basil line(SW) that produces phenylpropenes that are not para-O-methylated (Figure 4). The mRNA transcript expression ofphenylpropene OMTs (CVOMT and EOMT) followed thegeneral pattern of glandular trichome development on basilleaves. As the leaves begin to develop, they produce theglandular trichomes. As these develop, the enzyme levelsare high (as indicated by specific enzyme activity) and themRNA transcript levels are high as well. There also are moreglands per area in young leaves, before cell expansion(Gang et al., 2001). But as the glands reach maturity, thelevels of OMT enzyme activity and mRNA transcripts de-crease. These mRNA expression results supported enzymeassay results obtained previously (Wang, 1999; Lewinsohnet al., 2000; Gang et al., 2001). Quantitative RT-PCR estab-lished that �90 to 95% of the OMT transcripts encodedCVOMT, whereas the rest encoded EOMT. Because the cat-alytic efficiency of EOMT1 is �10-fold higher than the cata-lytic efficiency of CVOMT1, the relative proportions of thetwo transcript levels are consistent with the relative levels ofenzymatic activity observed in the plant (Figure 4).

Characterization of the recombinant CVOMT1 and EOMT1proteins demonstrated that CVOMT is specific for chavicol,whereas EOMT1 has similar levels of activity with chavicol asCVOMT but is much more active with eugenol as the substrate.These separate activities explain the differences in the ratios ofCVOMT and EOMT enzyme activities described above.

The difference in apparent Km values observed for bothCVOMT1 and EOMT1, as outlined in Table 1, must be inter-preted with caution because the two enzymes displayedsignificant substrate inhibition at �1 mM for their sub-strates. Thus, the activities of CVOMT1 and EOMT1 aremuch lower at chavicol concentrations of 5 mM than at 50�M. This property was observed for eugenol and t-isoeu-

genol as substrates for both EOMT1 and CVOMT1 as well,but it was strongest with chavicol as the substrate. Theseenzymes probably use an ordered substrate binding mecha-nism in which SAM binds first followed by the phenylpro-pene substrate. Substrate inhibition probably occursbecause S-adenosylhomocysteine (SAH) needs to be re-leased before the next phenylpropene substrate moleculebinds if a productive reaction complex is to form subse-quently. At higher phenylpropene substrate concentrations,however, the SAH is not released before another substratemolecule binds, leading to an unproductive complex. Thisproperty is facilitated by tight binding of SAH in the activesite, which probably binds almost as tightly as SAM (Zubietaet al., 2001), although this needs to be tested further.

It remains to be determined what the relative activities ofa heterodimeric enzyme (with one EOMT subunit and oneCVOMT subunit) with the two substrates would be, if suchheterodimers are formed. With a different set of OMT en-zymes, which are involved in alkaloid biosynthesis (Frickand Kutchan, 1999), such heterodimers—formed in a heter-ologous expression system—had novel substrate specifici-ties. However, the apparent Km values for the heterodimerswere several orders of magnitude higher that those for na-tive enzymes and endogenous substrates.

The fact that CVOMT1 has a lower catalytic efficiency forchavicol than EOMT1 has with eugenol should not lead tothe conclusion that CVOMT is not capable of serving as a vi-able enzyme in the plant glandular trichome system. On thecontrary, it is well established that enzymes involved in plantspecialized metabolism often have low turnover numbers.To overcome this lower catalytic efficiency, plants often ex-press such enzymes at very high levels in the appropriatetissue and cell type to achieve the required level of proteinactivity (White et al., 1998). It should be noted that EMX-1basil produces �20-fold more methylchavicol than methyl-eugenol (Gang et al., 2001), but this could not be the resultof differences in OMT activities (indeed, CVOMT and EOMTactivities in EMX-1 glands are comparable). The cause ofthe difference in methylchavicol and methyleugenol produc-tion in EMX-1 must reside in differences in earlier steps ofthe pathways leading to chavicol and eugenol.

Evolution of Enzyme Specificity in the SMOMT Family

The crystal structures of IOMT and ChOMT were reportedrecently (Zubieta et al., 2001). In addition, the structure ofCOMT also has been solved (C. Zubieta and J.P. Noel, un-published results), and this was found to be very similar tothe structures of IOMT and ChOMT. These structures re-vealed that the active sites of these enzymes are very similarin general structure. The central region of the active site islined with aromatic and aliphatic amino acid residues whosepositions and orientation appear to be responsible for pro-ducing the specific geometry that determines the level ofsubstrate discrimination that occurs with each enzyme. For

Basil Phenylpropene O-Methyltransferases 515

example, the side chains of Asn-310 and Cys-313 of IOMTwere found to be localized in the central region of the bind-ing pocket, and these were found to stabilize, via hydrogenbonds, the hydroxyl group of the proposed in vivo substrateof IOMT (Zubieta et al., 2001).

Molecular modeling of the CVOMT1 and EOMT1 aminoacid sequences, based on the IOMT and COMT crystal struc-tures, gave a rational explanation for the differences in sub-strate preference of these two closely related enzymes. Adifferent amino acid in a single position in the active site(Phe-260 in CVOMT1 and the equivalent Ser-261 in EOMT1)appeared to be responsible for the difference in substratespecificity between EOMT and CVOMT (Figures 6A and 6B).This explanation was tested empirically and shown to becorrect (Figure 7). In addition, the results of modeling of C.breweri IEMT also explain how the change of a single aminoacid in the progenitor COMT protein, changing Asn-134 toan alanine, also was sufficient to create a new enzyme(Wang and Pichersky, 1999) (Figure 6C).

The basil phenylpropene OMTs clearly evolved their sub-strate specificity independently of the analogous enzyme inC. breweri. This represents another example of the special-ized type of convergent evolution termed repeated evolution(Cseke et al., 1998; Pichersky and Gang, 2000). And al-though the evolution of basic phenylpropene OMT activity inthese two lineages may not be very recent, it is clear thatmore recent changes in enzyme specificity have occurred inthe basil lineage. However, it is not clear if CVOMT evolvedfrom EOMT or vice versa, and an analysis of related en-zymes in related species may be required to answer thisquestion. If CVOMT evolved from EOMT, then CVOMT evo-lution is a case of an enzyme evolving an increased sub-strate specificity and not a new substrate specificity. If thedirection was from CVOMT to EOMT, then this truly is acase of a new enzyme (i.e., with a new substrate specificity)that possibly originated very recently. Regardless of the ac-tual direction of evolution in this case, these results clearlyshow that a single amino acid substitution may be all that isrequired to produce a new SMOMT enzyme involved in plantspecialized (secondary) metabolite biosynthesis. A similardifference in substrate specificity attributable to a singleamino acid difference was reported by Frick and Kutchan(1999), although the actual in vivo substrates of the twoOMTs in that study were not clear. These results suggest thatnew enzymes in plant specialized metabolite biosynthesis(secondary metabolism) with new catalytic functions mayarise rapidly or instantaneously. Furthermore, because theevolution of new OMTs with new substrate specificities isrelatively simple, it is not surprising that OMTs with similarsubstrate specificities evolved independently in different plantlineages more than once.

Because of the great importance that such enzymes playin determining the defensive capabilities of the plant, inwhich new modifications of existing specialized compoundsneed to occur quickly in response to new pathogens or her-bivores or to the development of resistance in a present

pest, or to attract new pollinators, rapid evolution in the en-zymes involved may be the norm. Both gene duplication anddivergence and the faster route of divergence among allelesof loci encoding enzymes in plant specialized metabolite bio-synthesis (Pichersky and Gang, 2000) may be the means bywhich plants are able to respond to the continually changingenvironment in which they grow.

METHODS

Plant Material

Basil (Ocimum basilicum) lines EMX-1 and SW were grown undercontrolled conditions as described previously (Gang et al., 2001).

Chemicals and Radiochemicals

3H-Methyl-S-adenosyl-L-methionine (15 Ci/mmol) and 14C-methyl-S-adenosyl-L-methionine (55 mCi/mmol) were obtained from Amer-sham. S-Adenosyl-L-methionine and all other chemicals were pur-chased from Sigma Chemical Co. (St. Louis, MO), except the CoAesters, which were synthesized in the laboratory (see below).

Synthesis of CoA Esters

CoA esters of several hydroxycinnamic acids were synthesized usinga modified imidazolide method (Pabsch et al., 1991). Imidazolides ofthe corresponding acids were prepared as described previously(Pabsch et al., 1991). The reaction was monitored for purity and com-pletion by thin-layer chromatography (silica gel; solvent: diethyletheracidified with 1% acetic acid and visualized under UV light). Acid im-idazolides were used in twofold excess compared with the CoA so-dium salt (Sigma). The reaction was monitored as described abovewith a solvent system of n-butanol:acetic acid:water (63:10:27).CoA esters were identified with a delayed nitroprusside reaction(Stadtman, 1957). The reaction was terminated by extraction (threetimes) with ethyl acetate (phase separation achieved by centrifuga-tion). After evaporation of the organic solvent, ammonium acetatewas added to the water phase to a final concentration of 2% and themixture was loaded onto a preconditioned 1000-mg SPE cartridge(Chromabond C-18 ec; Machery-Nagel, Duren, Germany); condition-ing involved consecutive washes with methanol, dH2O, and 2% am-monium acetate solution (five column volumes each). The columnwas washed with 2% ammonium acetate solution until theflowthrough showed the absence of free CoA (determined with aspectrophotometer). The CoA esters were recovered by elution withdistilled water. Fractions containing the CoA esters were identifiedusing a spectrophotometer and lyophilized overnight. HPLC analysisshowed a purity of �90% (Nova-Pak RP-C18, 3.9 300 mm; Wa-ters, Milford, MA). Flow was 1 mL/min in solvent A (acetonitrile) andsolvent B (20 mM KH2PO4, pH 2.9). The gradient was 5% A in B for 0to 5 min, 5 to 38% A in B for min 5 to 32, 38 to 75% A in B for min 32to 35, 75 to 5% A in B for min 35 to 40, and 5% A in B for min 40 to45. CoA esters were stored lyophilized or in solution at pH 6.0 forseveral months at –80�C without noticeable degradation.

516 The Plant Cell

Cloning of Basil O-Methyltransferases

A cDNA for basil caffeic acid O-methyltransferase (COMT1) was iso-lated previously (Wang et al., 1999). An expressed sequence tag(EST) database was constructed previously by sequencing randomclones from a basil peltate glandular trichome cDNA library (Gang etal., 2001). These ESTs are available for searching upon request. Twotypes of closely related O-methyltransferase (OMT) genes were iden-tified in the EST database, although all sequences were incomplete.Rapid amplification of cDNA ends (Chenchik et al., 1996; Matz et al.,1999) and genome walking (Siebert et al., 1995) were used indepen-dently to obtain the sequence missing from the 5� end of the genes.Because the 5� and 3� ends of the genes were identical (the differ-ences were found in the middle of the coding region), two specificprimers were designed (5� primer, 5�-AATGGCATTGCAAAAAGTA-3�;3� primer, 5�-GATAGCTACAATTTTAAGG-3�) and used in reversetranscriptase–mediated polymerase chain reaction (RT-PCR) withthe basil peltate glandular trichome first-strand cDNA to obtain com-plete coding regions. Two unique genes (CVOMT1 for chavicol OMTand EOMT1 for eugenol OMT) were amplified and transferred intothe pCRT7/CT-TOPO TA expression vector (Invitrogen). Aftercomplete sequencing, the resulting constructs were transformedinto Escherichia coli BL21(DE3)pLysS cells for expression.

Sequence Analysis

Amino acid sequence alignments and neighbor-joining trees weregenerated using the ClustalX computer program (Thompson et al.,1997). The resulting dendrograms were visualized using the NJplotcomputer program (Perrière and Gouy, 1996).

RNA Gel Blot Analysis

Total RNAs from isolated peltate glandular trichomes and from wholeleaf tissue from basil lines EMX-1 and SW were purified as describedpreviously (Gang et al., 2001). A 32P-labeled probe for CVOMT1 wasconstructed using the Rediprime II Kit (Amersham Pharmacia Bio-tech) from a gel-purified PCR product (�500 bp) amplified usingprimers designed from the CVOMT1 cDNA sequence (CVOMT1 5�

probe primer, 5�-TTTCCCAATTACTCAAGGCC-3�; CVOMT1 3� probeprimer, 5�-CCCTCCAACGCCAACTGC-3�). Formaldehyde agarose(1%) gel electrophoresis, RNA gel blot analysis, probe hybridizationat 65�C, and washes were performed under standard conditions(Sambrook et al., 1989) using Hybond-XL nylon membranes (Amer-sham Pharmacia Biotech) and 4 SSC (1 SSC is 0.15 M NaCl and0.015 M sodium citrate) as the hybridization buffer.

Quantitative RT-PCR

Total RNAs, isolated as for RNA gel blot analysis, from 0.5-, 1-, and3-cm leaves as well as from peltate glands were used with the First-Strand cDNA Synthesis Kit (Amersham Pharmacia Biotech) to pro-duce first-strand cDNAs. Gene-specific primer pairs were designedthat amplified only the gene of interest when control PCR procedureswere performed; these reactions used plasmids containing eitherCVOMT1 or EOMT1 as a template. The primer pairs were CVOMT15� probe primer with CVOMT1 3� probe primer and EOMT1 5� probeprimer (5�-TTTCCCAGTTACTCCAATCT-3�) with EOMT1 3� probeprimer (5�-CCCTTCCATAACGACCGT-3�). Several amplification cy-

cle numbers were tested for each cDNA sample. The optimal numberof cycles, which produced faint yet clearly detectable bands for theCVOMT1 gene products, was determined to be 23 cycles for 1-cmleaf and peltate gland cDNAs and 26 cycles for 0.5- and 3-cm leafcDNAs. These cycle numbers were used for templates with CVOMT1and EOMT1 primer sets in triplicate 50-�L PCR procedures. An iden-tical number of cycles was used for both CVOMT1 and EOMT1 foreach leaf cDNA sample. Slot blotting and DNA gels followed by DNAgel blotting were then performed in duplicate with 10 �L of eachsample. Hybond N� nylon membrane (Amersham Pharmacia Bio-tech) was used for both applications. The resulting blot pairs thenwere prehybridized and hybridized (according to the manufacturer’sinstructions) to either CVOMT1- or EOMT1-labeled probe, producedfrom PCR products (using the appropriate primer pair with the corre-sponding plasmid templates), which were labeled using the Re-diprime II kit (Amersham Pharmacia Biotech). After overnightincubation at 65�C, the blots were washed according to the mem-brane manufacturer’s instructions. After overnight exposure to aphosphorimager screen (Bio-Rad Laboratories), the relative signal in-tensity for each sample was measured using the Molecular Imagersoftware package (Bio-Rad Laboratories). DNA gel blots and slotblots gave comparable results.

Preparation of Crude Cell-Free Extracts from E. coli

Single isolated bacterial colonies from freshly streaked plates (grownon Luria-Bertani agar medium containing 50 �g/mL ampicillin and 15�g/mL chloramphenicol) were used to inoculate 2-mL liquid cultures(in Luria-Bertani medium with 50 �g/mL ampicillin), which weregrown overnight at 37�C. An aliquot (500 �L) of these cultures wasused to inoculate 50-mL liquid cultures. Once the cultures reached acell density of 0.3 to 0.6 OD600, recombinant protein expression wasinduced by the addition of 0.3 mM isopropylthio-�-galactoside. Afterovernight incubation at room temperature, the cells were pelleted bycentrifugation at 20,000g for 10 min at 4�C. The pellets were storedat 20�C. Cells containing expressed recombinant protein were re-suspended in 500 �L of 50 mM bis-Tris, pH 7.0, containing 10%glycerol, 14 mM 2-mercaptoethanol, 1 mM EDTA, and 10 mM NaCl.After the cells were lysed by adding lysozyme to a final concentrationof 1 �g/mL, incubating the resulting mixture on ice for 20 min, andfreezing (in liquid N2) and thawing two times, cellular debris were re-moved by centrifugation at 20,000g for 10 min at 4�C. This crudebacterial lysate was used for further purification steps and enzymaticassays.

Partial Enzyme Purification

All procedures were performed at 0 to 4�C using buffer A (50 mM bis-Tris, pH 7.0, containing 10% glycerol, 5 mM Na2S2O5, 10 mM 2-mer-captoethanol, and 1 mM EDTA). Protein concentrations were mea-sured according to Bradford (1976) using the Bio-Rad proteinreagent with BSA (Sigma) as a standard.

Gel Filtration

The crude bacterial lysate (3 mL) was applied to a Bio-Gel P-6 col-umn (Bio-Rad, Hercules, CA; 1 10 cm) and preequilibrated inbuffer A, and proteins were eluted in the same buffer. Eluting frac-tions containing enzyme activity were combined.

Basil Phenylpropene O-Methyltransferases 517

Ion Exchange Chromatography

The pooled fractions (�2 mL) were applied to a Hitrap Q Sepharosecolumn (1 5 cm; Amersham Pharmacia Biotech) equilibrated withbuffer A. The flow rate was 1 mL/min. Buffer B was the same asbuffer A but contained 1 M KCl. Proteins were eluted from the col-umn with the following gradient program: 10 mL of 0% buffer B, 40mL of 0 to 60% buffer B, and 10 mL of 100% buffer B. Fractions weretested for enzyme activity, and active fractions were combined.

Subunit and Native Molecular Mass

The subunit molecular mass of each enzyme was determined bySDS-PAGE (12% polyacrylamide), and the molecular mass of the na-tive enzymes was estimated using gel filtration chromatographythrough a Superdex 75 Hiload Prep 16/60 column (Amersham Phar-macia Biotech). Molecular mass standards were alcohol dehydroge-nase (150 kD), phosphorylase b (97 kD), BSA (66 kD), carbonicanhydrase (29 kD), cytochrome c (12.4 kD), and aprotinin (6.5 kD), allfrom Sigma.

Enzyme Assays

The standard radiochemical assay mixture for the recombinant pro-teins consisted of 70 �L of buffer A, 10 �L of enzyme solution (1.5 to3 �g of protein), 10 �L of substrate solution (in ethanol; final concen-tration of substrates was 1 mM for COMT and 0.1 mM for CVOMTand EOMT), and 10 �L of 3H-S-adenosylmethionine (SAM) (20,000cpm per reaction; final concentration of substrates was 180 �M forCOMT and 15 �M for CVOMT and EOMT) in a total volume of 100�L. After incubation at 30�C for 0.5 hr, the reactions were stopped bythe addition of 10 �L of 2 N HCl. Radiolabeled products then wereextracted by the addition of 1 mL of ethyl acetate followed by vortex-ing. After centrifugation (1 min at 20,000g) to separate the phases,800 �L of the upper (ethyl acetate) phase was used for scintillationcounting (3 mL of scintillation fluid—4 g/L 2,5-diphenyloxazol and0.05 g/L 2,2�-p-phenylene-bis[5-phenyloxazol] in toluene—in a liquidscintillation counter). Verification of assay product identity wasachieved by performing assays as described above but using nonra-diolabeled SAM as cofactor and concentrating the ethyl acetate–extracted products under dry nitrogen. The concentrated productsthen were resuspended in an appropriate solvent and analyzed bygas chromatography–mass spectrometry (GC-MS) or liquid chroma-tography–mass spectrometry (LC-MS).

GC-MS Analysis

Volatile compounds were analyzed with a Hewlett-Packard GC-MSsystem (Palo Alto, CA) equipped with an HP-5 [30 m 0.25 mm]fused silica capillary column. Helium (1 mL/min) was used as a carriergas. The injector was set for splitless injection at 250�C, and the de-tector was set at 280�C. The oven was set to 50�C for 1 min after in-jection, and then the temperature was increased to 200�C at a rate of4�C/min. The mass-to-charge ratio was 45 to 450, with an energy of70 electron volts. Eluting compounds were identified using a GC-MSlibrary and by comparison of mass spectra and retention times withauthentic samples.

LC-MS Analysis

Nonvolatile compounds were analyzed on a Micromass Quattro LCZtriple-quadropole mass spectrometer (Micromass, Beverly, MA) at-tached to a Waters 2690 Separations Module with attached columnoven. HPLC separation of the compounds over a Waters Nova-PakC18 column (4.6 mm 30 cm) was achieved using a 30-min lineargradient from 5% acetonitrile in 0.05% formic acid to 100% acetoni-trile, with the flow rate set at 1 mL/min and the column temperatureset to 40�C. In-line UV light spectra (200 to 600 nm) were obtainedusing an attached Waters 996 photodiode array detector. Splittingbefore entrance into the mass spectrometer reduced the flow rate to100 �L/min. Electrospray ionization in the Micromass Z-Spraysource was achieved in negative ion mode by setting the capillaryvoltage to 2.5 to 3.0 kV and setting the cone voltage to between 30and 80 V depending on the compound to be analyzed. The desolva-tion gas flow rate and temperature were set to 270 L/hr and 250�C,respectively, and the source temperature was set to 140�C. Othersource parameters were set to standard conditions recommendedby Micromass. Eluting compounds were identified by comparison ofUV light spectra, mass spectra, and elution volume with authenticsamples (analyzed on the same instrument with identical HPLC elu-tion and source parameters).

Kinetic Properties

Standard methods (e.g., using Lineweaver-Burk, Eddie-Hofstee, andMichaelis-Menten equations [Segel, 1993]) were used to determineapparent Km and Vmax for CVOMT, EOMT, and COMT from basil us-ing modified standard radiochemical assays. For apparent Km withSAM, the concentration of substrate (chavicol for CVOMT, eugenolfor EOMT, and caffeic acid for COMT) was kept constant at 0.1, 0.1,and 1 mM, respectively, whereas the concentration of SAM was var-ied. For apparent Km with various substrates, SAM concentration waskept constant at 180 �M for COMT and 15 �M for CVOMT and EOMT,whereas the concentrations of the other substrates were varied.

Molecular Modeling

The sequences for CVOMT, EOMT, and (iso)eugenol OMT were fit to thethree-dimensional structure obtained for isoflavone OMT using Modeller(Sali and Blundell, 1993). The resulting pdb files were visualized usingSwiss PDB Viewer (Guex and Peitsch, 1997), and three-dimensionalstereo images were drawn using POV-Ray (X-POV-Team, 1997).

Construction of Mutant OMT Enzymes

Ser-261 in EOMT was mutated to a phenylalanine residue and Phe-260 in CVOMT was mutated to a serine residue via the DNA PCRmethod (Ho et al., 1989) using Taq DNA polymerase (Fisher Scien-tific). The external primers were CVOMT-Nt1 (5�-AATGGCATT-GCAAAATATG-3�) and OMT-Ct1 (5�-TTTTAAGGATAAGCCTCTA-3�)for CVOMT1 and EOMT-Nt1 (5�-AATGGCATTGCAAAAAGTA-3�) andOMT-Ct1 (5�-TTTTAAGGATAAGCCTCTA-3�) for EOMT1. One set ofcomplementary mutating primers was designed for each gene:CVOMT1-F1 (5�-CTTCTCAAGTCTATAATACACGA-3�) and CVOMT1-R1(5�-TCGTGTATTATAGACTTGAGAAG-3�) for CVOMT1 and EOMT1-F1(5�-CTTCTAAAGTTTATAATACATGA-3�) and EOMT1-R1 (5�-TCATGT-ATTATAAACTTTAGAAG-3�) for EOMT1. Bases coding for the mutated

518 The Plant Cell

amino acid are shown in italics, and bases causing the mutation areshown in boldface.

Accession Numbers

The accession numbers for the proteins listed in Figure 2 areAAD24001 (Pinus radiata COMT), AAC49708 (Pinus taeda COMT),AAB71141 (Clarkia breweri COMT), AAC01533 (C. breweri IEMT),CAA52814 (Eucalyptus gunnii COMT), AAB46623 (Medicago sativaCOMT), CAA01820 (Populus balsamifera COMT), CAA58218 (Prunusdulcis COMT), AAB96879 (Arabidopsis thaliana COMT1), AAA86982(Chrysosplenium americanum COMT), AAD38189 (Ocimum basili-cum COMT1), S36403 (Nicotiana tabacum COMT), T12259 (Capsicumannuum OMT), AAA86718 (Zinnia elegans COMT), AAB48059 (M. sativaChOMT), T04963 (A. thaliana OMT), S52015 (Hordeum vulgare F7OMT),P47917 (Zea mays OMT), AAD10485 (Triticum aestivum OMT), T06786(Pisum sativum HMOMT), T09299 (M. sativa OMT), T09254 (M. sativaIOMT), AF435007 (O. basilicum CVOMT1), AF435008 (O. basilicumEOMT1), BAA86059 (Pyrus pyrifolia OMT), CAA11131 (P. dulcis OMT),and AAB71213 (Prunus armeniaca OMT).

ACKNOWLEDGMENTS

This research was supported in part by a competitive grant from theUnited States Department of Agriculture National Research Initiative(Grant No. 0003497). T.B. was supported in part by a Deutscher Ak-ademischer Austausch Dienst fellowship (Gemeinsames Hochschul-sonderprogramm III von Bund und Ländern, Germany).

Received August 2, 2001; accepted November 1, 2001.

REFERENCES

Adams, S., and Weidenborner, M. (1996). Mycelial deformations ofCladosporium herbarum due to the application of eugenol or car-vacrol. J. Essential Oil Res. 8, 535–540.

Bradford, M.M. (1976). A rapid and sensitive method for the quanti-tation of microgram quantities of protein utilizing the principle ofprotein-dye binding. Anal. Biochem. 72, 248–254.

Chatterjee, A., Sukul, N., Laskar, S., and Ghoshmajumdar, S.(1982). Nematicidal principles from two species of Lamiaceae, Oci-mum sanctum and Ocimum basilicum. J. Nematol. 14, 118–122.

Chenchik, A., Zhu, Y., Diatchenko, L., Li, R., Hill, J., and Siebert,P. (1996). Generation and use of high-quality cDNA from smallamounts of total RNA by SMART PCR. In RT-PCR Methods forGene Cloning and Analysis, P. Siebert and J. Larrick, eds (West-borough, MA: BioTechniques Books), pp. 305–319.

Collendavelloo, J., Legrand, M., Geoffroy, P., Barthelemy, J., andFritig, B. (1981). Purification and properties of the three o-diphenol-O-methyltransferases of tobacco leaves. Phytochemistry 20, 611–616.

Cseke, L., Dudareva, N., and Pichersky, E. (1998). Structure andevolution of linalool synthase. Mol. Biol. Evol. 15, 1491–1498.

Frick, S., and Kutchan, T.M. (1999). Molecular cloning and functional

expression of O-methyltransferases common to isoquinoline alka-loid and phenylpropanoid biosynthesis. Plant J. 17, 329–339.

Gang, D.R., Wang, J., Dudareva, N., Nam, K.H., Simon, J.,Lewinsohn, E., and Pichersky, E. (2001). An investigation of thestorage and biosynthesis of phenylpropenes in sweet basil (Oci-mum basilicum L.). Plant Physiol. 125, 539–555.

Gildemeister, E., and Hoffmann, F.R. (1913). Ätherischen Öle. (Mil-waukee, WI: Pharmaceutical Review Publishing Co.).

Grossman, J. (1993). Botanical pesticides in Africa. Int. PestManag. Pract. 15, 1–9.

Guenther, E. (1949). The Essential Oils. (Princeton, NJ: D. Van Nos-trand Co., Inc.).

Guex, N., and Peitsch, M.C. (1997). SWISS-MODEL and the Swiss-PdbViewer: An environment for comparative protein modeling.Electrophoresis 18, 2714–2723.

Guo, D.J., Chen, F., Inoue, K., Blount, J.W., and Dixon, R.A.(2001). Downregulation of caffeic acid 3-O-methyltransferase andcaffeoyl CoA 3-O-methyltransferase in transgenic alfalfa: Impactson lignin structure and implications for the biosynthesis of G andS lignin. Plant Cell 13, 73–88.

He, X.-Z., Reddy, J.T., and Dixon, R.A. (1998). Stress responses inalfalfa (Medicago sativa L.). XXII. cDNA cloning and characteriza-tion of an elicitor-inducible isoflavone 7-O-methyltransferase.Plant Mol. Biol. 36, 43–54.

Ho, S.N., Hunt, H.D., Horton, R.M., Pullen, J.K., and Pease, L.R.(1989). Site-directed mutagenesis by overlap extension using thepolymerase chain reaction. Gene 77, 51–59.

Ibrahim, R.K., Bruneau, A., and Bantignies, B. (1998). Plant O-meth-yltransferases: Molecular analysis, common signature and classifi-cation. Plant Mol. Biol. 36, 1–10.

Karapinar, M., and Aktug, S. (1987). Inhibition of foodborne patho-gens by thymol, eugenol, menthol and anethole. Int. J. FoodMicrobiol. 4, 161–166.

Klischies, M., Stockigt, J., and Zenk, M.H. (1975). Biosynthesis ofthe allylphenols eugenol and methyleugenol in Ocimum basilicumL. Chem. Commun. 21, 879–880.

Lawrence, B.M. (1992). Chemical components of labiate oils andtheir exploitation. In Advances in Labiate Science, R.M. Harleyand T. Reynolds, eds (Kew, UK: Royal Botanical Gardens).

Lewinsohn, E., Ziv-Raz, I., Dudai, N., Tadmor, Y., Lastochkin, E.,Larkov, O., Chaimovitsh, D., Ravid, U., Putievsky, E., Pichersky,E., and Shoham, Y. (2000). Biosynthesis of estragole and methyl-eugenol in sweet basil (Ocimum basilicum L): Developmental andchemotypic association of allylphenol O-methyltransferase activi-ties. Plant Sci. 160, 27–35.

Manitto, P., Monti, D., and Gramatica, P. (1974). Biosynthesis ofphenylpropanoid compounds. I. Biosynthesis of eugenol in Oci-mum basilicum L. J. Chem. Soc. Perkin Trans. I 14, 1727–1731.

Manitto, P., Gramatica, P., and Monti, D. (1975). Biosynthesis ofphenylpropanoid compounds. II. Incorporation of specificallylabelled cinnamic acids into eugenol in basil [Ocimum basilicum].J. Chem. Soc. Perkin Trans. I 16, 1549–1551.

Martz, F., Maury, S., Pincon, G., and Legrand, M. (1998). cDNAcloning, substrate specificity and expression study of tobaccocaffeoyl-CoA 3-O-methyltransferase, a lignin biosynthetic enzyme.Plant Mol. Biol. 36, 427–437.

Basil Phenylpropene O-Methyltransferases 519

Matz, M., Lukyanov, S., Bogdanova, E., Diatchenko, L., andChenchik, A. (1999). Amplification of cDNA ends based on tem-plate-switching effect and step-out PCR. Nucleic Acids Res. 27,1558–1560.

Maury, S., Geoffroy, P., and Legrand, M. (1999). Tobacco O-meth-yltransferases involved in phenylpropanoid metabolism: The dif-ferent caffeoyl-coenzyme A/5-hydroxyferuloyl-coenzyme A 3/5-O-methyltransferase and caffeic acid/5-hydroxyferulic acid 3/5-O-methyltransferase classes have distinct substrate specificitiesand expression patterns. Plant Physiol. 121, 215–223.

Maxwell, C.A., Edwards, R., and Dixon, R.A. (1992). Identification,purification, and characterization of S-adenosyl-L-methionine-isoliquiritigenin 2�-O-methyltransferase from alfalfa (Medicagosativa L). Arch. Biochem. Biophys. 293, 158–166.

Maxwell, C.A., Harrison, M.J., and Dixon, R.A. (1993). Molecularcharacterization and expression of alfalfa isoliquiritigenin 2�-O-methyltransferase, an enzyme specifically involved in the bio-synthesis of an inducer of Rhizobium meliloti nodulation genes.Plant J. 4, 971–981.

Mbeguie-A-Mbeguie, D., Gomez, R.-M., and Fils-Lycaon, B.(1997). Sequence of an O-methyltransferase from apricot fruit(accession no. U82011): Gene expression during fruit ripening(PGR97–118). Plant Physiol. 114, 1569.

Miyao, S. (1975). Inhibitory effects of ethanol extract of mace andeugenol on the growth of microorganisms isolated from Viennasausages. Nihon Shokuhin Eisei Gakkai 16, 412–416.

Obeng-Ofori, D., and Reichmuth, C. (1997). Bioactivity of eugenol,a major component of essential oil of Ocimum suave (Wild.)against four species of stored-product Coleoptera. Int. J. PestManage. 43, 89–94.

Pabsch, K., Petersen, M., Rao, N.N., Alfermann, A.W., and Wandrey,C. (1991). Chemo-enzymatic synthesis of rosmarinic acid. Rec. Trav.Chim. Pays-Bas 110, 199–205.

Parvathi, K., Chen, F., Guo, D.J., Blount, J.W., and Dixon, R.A. (2001).Substrate preferences of O-methyltransferases in alfalfa suggest newpathways for 3-O-methylation of monolignols. Plant J. 25, 193–202.

Pellegrini, L., Geoffroy, P., Fritig, B., and Legrand, M. (1993). Molec-ular cloning and expression of a new class of ortho-diphenol-O-meth-yltransferases induced in tobacco (Nicotiana tabacum L.) leaves byinfection or elicitor treatment. Plant Physiol. 103, 509–517.

Perrière, G., and Gouy, M. (1996). WWW-Query: An on-line retrievalsystem for biological sequence banks. Biochimie 78, 364–369.

Pichersky, E., and Gang, D.R. (2000). Genetics and biochemistry ofsecondary metabolites in plants: An evolutionary perspective.Trends Plant Sci. 5, 439–445.

Sali, A., and Blundell, T.L. (1993). Comparative protein modellingby satisfaction of spatial restraints. J. Mol. Biol. 234, 779–815.

Sambrook, J., Fritsch, E.F., and Maniatis, T. (1989). MolecularCloning: A Laboratory Manual, 2nd ed. (Cold Spring Harbor, NY:Cold Spring Harbor Laboratory Press).

Sangwan, N., Verman, B., Verma, K., and Dhindsa, K. (1990).Nematicidal activity of some essential plant oils. Pestic. Sci. 28,331–335.

Schmitt, D., Pakusch, A.E., and Matern, U. (1991). Molecular-clon-ing, induction, and taxonomic distribution of caffeoyl-CoA 3-O-meth-yltransferase, an enzyme involved in disease resistance. J. Biol.Chem. 266, 17416–17423.

Segel, I.H. (1993). Enzyme Kinetics: Behavior and Analysis of RapidEquilibrium and Steady-State Enzyme Systems. (New York: JohnWiley and Sons).

Senanayake, U.M., Wills, R.B.H., and Lee, T.H. (1977). Biosynthe-sis of eugenol and cinnamic aldehyde in Cinnamomum zeylani-cum. Phytochemistry 16, 2032–2033.

Shukla, R., and Prasad, V. (1985). Population fluctuations of theoriental fruit fly, Dacus dorsalis Hendel, in relation to hosts andabiotic factors. Trop. Pest Manage. 31, 273–275.

Siebert, P.D., Chenchik, A., Kellogg, D.E., Lukyanov, K.A., andLukyanov, S.A. (1995). An improved method for walking inuncloned genomic DNA. Nucleic Acids Res. 23, 1087–1088.

Sisk, C., Shorey, H., Gerber, R., and Gaston, L. (1996). Semio-chemicals that disrupt foraging by the Argentine ant (Hymenoptera:Formicidae): Laboratory bioassays. J. Econ. Entomol. 89, 381–385.

Stadtman, E.R. (1957). Preparation and Assay of Acyl Coenzyme Aand Other Thiol Esters: Use of Hydroxylamine. Methods Enzymol.III, 931–941.

Suelves, M., and Puigdomenech, P. (1998). Specific mRNA accu-mulation of a gene coding for an O-methyltransferase in almond(Prunus amygdalus, Batsch) flower tissues. Plant Sci. 134, 79–88.

Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., andHiggins, D.G. (1997). The ClustalX Windows interface: Flexiblestrategies for multiple sequence alignment aided by quality analy-sis tools. Nucleic Acids Res. 24, 4876–4882.

Vander Mijnsbrugge, K., Meyermans, H., Van Montagu, M.,Bauw, G., and Boerjan, W. (2000). Wood formation in poplar:Identification, characterization, and seasonal variation of xylemproteins. Planta 210, 589–598.

Wang, J. (1999). Molecular Characterization of an O-Methyltrans-ferase Involved in Floral Scent Production in Clarkia breweri. Ph.D.Dissertation. (Ann Arbor, MI: University of Michigan).

Wang, J., and Pichersky, E. (1998). Characterization of S-adenosyl-L-methionine:(iso)eugenol O-methyltransferase involved in scent pro-duction in Clarkia breweri. Arch. Biochem. Biophys. 349, 153–160.

Wang, J., and Pichersky, E. (1999). Identification of specific resi-dues involved in substrate discrimination in two plant O-methyl-transferases. Arch. Biochem. Biophys. 368, 172–180.

Wang, J., Dudareva, N., Kish, C.M., Simon, J.E., Lewinsohn, E.,and Pichersky, E. (1999). Nucleotide sequences of two cDNAsencoding caffeic acid O-methyltransferases from sweet basil (Oci-mum basilicum). Plant Physiol. 120, 1205.

White, W.L.B., Arias-Garzon, D.I., McMahon, J.M., and Sayre,R.T. (1998). Cyanogenesis in cassava: The role of hydroxynitrilelyase in root cyanide production. Plant Physiol. 116, 1219–1225.

Wu, Q., Preisig, C.L., and VanEtten, H.D. (1997). Isolation of thecDNAs encoding (�)-6a-hydroxymaackiain 3-O-methyltransferase,the terminal step for the synthesis of the phytoalexin pisatin inPisum sativum. Plant Mol. Biol. 35, 551–560.

X-POV-Team. (1997). POV-Ray: Persistence of vision ray-tracer.Available at http://www.povray.org.

Ye, Z.-H., and Varner, J. (1995). Differential expression of twoO-methyltransferases in lignin biosynthesis in Zinnia elegans.Plant Physiol. 108, 459–467.

Zubieta, C., He, X.-Z., Dixon, R.A., and Noel, J.P. (2001). Structures oftwo natural product methyltransferases reveal the basis for substratespecificity in plant O-methyltransferases. Nat. Struct. Biol. 8, 271–279.